The challenging end-to-end pre-flight calibration of FIREBall-2 at the launch base

The NASA/CNES co-funded instrument is a balloon-borne 1-m telescope coupled to a UV (200 nm) multi-object spectrograph designed to image the circum-galactic medium (CGM) in emission from the stratosphere (~40km, 3mB). FIREBall-2 shows a wide field of view (FOV) of 20.5x37 arcmin2, a high speed beam (f/2.5) and a spatial and spectral resolution of respectively ∼5 arcsec FWHM, and ∼2000 (λ/dλ). Nine science or calibration masks can be selected. A guider camera in the visible uses the full pupil flux at field locations not used for UV targets. A UV multi-object spectrograph is, by its very nature, at the limit of complexity of what can be calibrated at a launch base with extremely limited GSE (Ground Support Equipment). Indeed, a major challenge of the FIREBall-2 project was to develop a 100% self-consistent method that did not require any GSE for all the pre-flight adjustments and calibration or a space facility. To this end, invaluable assets of FIREBall-2 are its siderostat which provides a full-pupil auto-collimation capability, and the good atmospheric transmission at 200 nm up to distances of a few tens meters. The auto-collimation mode has been used extensively in the calibration process, in particular to: * solve the very demanding plate scale verification and absolute X, Y in-flight positioning of the targets into the 6 arc seconds slits (~80 μm), * achieve and assess a 5 arcseconds image quality over the 0.6 degree UV FOV with a f/2.5 aperture ratio very sensitive to focus changes. The auto-collimation has been completed with on-sky observation in the visible of all target fields for guidance testing and ground reference of infinity in the guider camera. A ZEMAX-coupled instrument model developed at LAM under Python has been used to support this calibration (distortion - dispersion - throughfocus analysis). This model has also been used to make an end-to-end prediction of the observations of the CGM emission from a large halo in a cosmological simulation (Ramona et al. in prep).